Chaotic Time Series Analysis With Application To The Study On The Foreign Exchange Market

With the rapid development of China's financial market, prediction analysis theory isextensively applied in domestic investment community. After the entry into the WTO, importand export trade are developing steadfastly. The foreign exchange market plays a significantrole in this field, and is the focus of attention.The foreign exchange market are controlled by various factors, its presentation form is fea-tured by both definitude and randomicity. Therefore, it is necessary to use some specificmethods to conduct prediction. These data are considered as time series, and modeling andpredicting are conducted with the application of time series analysis.This thesis studies the application of chaotic time series in the foreign exchange mar-ket. It gives an introduction on the identification of chaos, the selection of phase spacereconstruction parameters of chaotic time series, and the predicting methods of chaotic timeseries. And also, these methods are utilized in predicting of USD exchange rate with thechaotic model. The main contents of the article are as follows:First, the development of nonlinear science such as chaotic science in today's societyis introduced; the background information and research status of chaotic time series predicttheory are summarized; the existing issues in current chaotic time series predict research areanalyzed.Second, identification and the diagnosis of chaotic system and some commonly appliedpredicting methods, such as regional linearity predict, nonlinear multi-step prediction, andRBF prediction are analyzed. Next, monovariate time series predicting method and multivariate time series predictingmethod are comparatively analyzed. With the same condition , and via numerical simulation,it is apparent that multivariate time series can obtain better predict accuracy.Finally, we apply the chaotic time series prediction theory to study the change of USD.Then by reconstruction systematic equation, and the integration of embed theory, data pass-ing nonlinear region after denoising treatment are predicted in multiple steps. And these dataare compared with actual data and shows the good predicted result are obtained.